Effect of Metformin on Renal Function After Primary Percutaneous Coronary Intervention in Patients Without Diabetes Presenting with ST-elevation Myocardial Infarction: Data from the GIPS-III Trial

Purpose The association between metformin use and renal function needs further to be elucidated since data are insufficient whether metformin affects renal function in higher risk populations such as after ST-elevation myocardial infarction (STEMI). Methods We studied 379 patients included in the GIPS-III trial in which patients without diabetes or renal dysfunction, who underwent primary percutaneous coronary interventions (PCI) for STEMI, were randomized to metformin 500 mg or placebo twice daily for four months. At baseline and at seven scheduled visits up to four months after PCI, estimated glomerular filtration rate (eGFR) was determined (2582 values). Contrast-induced acute kidney injury (CI-AKI) was defined as an increase in serum creatinine of ≥0.3 mg/dl or 25 % rise within 48 h after PCI. Results At all visits, the mean eGFR was similar in patients randomized to metformin or placebo. Over the four month period, mixed-effect repeated-measures model analysis showed a least-squares mean ± standard error change in eGFR of -5.9 ± 0.8 ml/min/1.73 m2 in the metformin group and −7.1 ± 0.8 ml/min/1.73 m2 in the control group (P = 0.27 for overall interaction). The incidence of CI-AKI was 14.8 %; 29 (15.2 %) patients in the metformin group versus 27 (14.4 %) controls (P = 0.89). After adjustment for covariates, metformin treatment was not associated with CI-AKI (odds ratio: 0.96, 95%CI 0.52 − 1.75, P = 0.88). Conclusion We conclude that initiation of metformin shortly after primary PCI has no adverse effect on renal function in patients without diabetes or prior renal impairment, further providing evidence of the safety of metformin use after myocardial infarction and subsequent contrast exposure. Electronic supplementary material The online version of this article (doi:10.1007/s10557-015-6618-1) contains supplementary material, which is available to authorized users.


Assessment of renal function
The measurements of serum creatinine concentrations were planned at baseline, 6h, 12h, 24h, 48h, 2 weeks, 6 weeks, and 4 months after PCI. 2 For patients who were returned to their a referring center after PCI, creatinine concentration measurements during hospital admission and up to four months after PCI were obtained. When more than one measurement was available at a specific time point, the measurement that was nearest to the designated time point was used for analysis.
Estimated glomerular filtration rate (eGFR) was calculated using the Chronic Kidney Disease Epidemiology Collaboration study equation 4 , as has been suggested to estimate the GFR more accurate than the simplified Modification of Diet in Renal Disease formula in patients without severe renal impairment. 5 Contrast-induced acute kidney injury was defined as an increase in serum creatinine of ≥0.3 mg/dL (27 μmol/L), or a 25% relative rise in creatinine, within 48 hours after the start of the PCI procedure. 6 Time of lidocaine injection defined the start of the PCI procedure. If unavailable, arrival time at the catheterization lab was used. The application of a closure device defined the end of the procedure. The extent of myocardial damage was estimated by calculating the area under the curve of CK and the myocardial band of CK from baseline until 24 hours after PCI.

Statistical analysis
Normally distributed continuous data are presented as means with standard error, unless stated otherwise, and differences were assessed using the Student's t-test. If normality could not be assumed, the data are presented as medians with interquartile range, and differences were assessed using the Mann-Whitney U test. Categorical data are presented as frequencies with percentages, and differences were assessed using the chi-square or the Fisher exact test, when appropriate.
To estimate and compare the effect of metformin on renal function, mixed-effects repeated measures analysis with random slope and intercept was performed. This type of analysis operates using estimation techniques that allow for incomplete data. Age, gender, baseline NT-proBNP concentration, and myocardial blush grade after PCI were, based on the statistical plan of the main study, considered as covariables in relation to the renal function. 1 Only significant covariables were used as fixed effects in the final multivariate model. Individual patients were considered as a random effect. The covariance matrix of residuals used in the model was unstructured.
The independent predictors of CI-AKI were identified by a backward-stepwise logistic regression model using an entry level of significance of 0.1. Randomization to metformin or placebo, gender, age, contrast dose, eGFR at baseline, and anemia (hemoglobin (Hb) <13.7 mg/dl (<8.5 mmol/L) in men and Hb <12.1 mg/dl (<7.5 mmol/L) in women) were previously associated with the development of CI-AKI and were therefore forced in the multivariate analysis in addition to risk factors found in the univariate analysis. [7][8][9][10] All reported P-values are 2-sided and P<0.05 was considered statistically significant, except for interactions in which P<0.10 was considered significant. Since this study concerns a post-hoc analysis of a previously conducted randomized controlled trial, no sample size calculation was made. Statistical analyses were performed by using SPSS Statistics for Windows, Version 22 (IBM, Armonk, NY) and Stata version 12.0 (Stata Corp., College Station, TX).  .26 Values are expressed as mean (standard error of the mean). Estimated glomerular filtration rate (eGFR) was adjusted using a mixed-effects repeated measurements model for age, baseline Nterminal pro-B-type natriuretic peptide concentration, and treatment allocation. We assumed an unstructured covariance structure among serial observations. * Calculated using the Chronic Kidney Disease Epidemiology Collaboration study equation. ** After adjustment for baseline eGFR.

Characteristic
† Determined using the Student's t-test.

Online Resource 5 -Unadjusted renal function and differences in renal function at individual time points
Values are expressed as mean (standard error of the mean). eGFR estimated glomerular filtration rate. Values are expressed as no (%). ARB angiotensin-receptor blocker.
Online Resource 8 -Presented are least-squares means ± standard error from the mixed-effects repeated measurements model with a random intercept and slope. Individual patients were considered as random effects and the following were fixed effects: age, baseline N-terminal pro-B-type natriuretic peptide concentration, development of CI-AKI, and treatment allocation. We assumed an unstructured covariance structure among serial estimated glomerular filtration rate values (eGFR). No significant difference was observed for the overall interaction of CI-AKI, time and allocated treatment (P=0.14). There was an expected overall interaction between the occurrence of CI-AKI and the change in eGFR over the first 48 hours (P=0.021), but there was no difference between randomized treatments (P=0.13). At the end of follow up, eGFR did not differ between patients on placebo or metformin, neither in the CI-AKI group (P=0.34) nor the no CI-AKI group (P=0.66).